Mat. Res. Soc. Symp. Proc. Vol. 644 200 Materials Research Society ELASTIC PROPERTIES OF THREE BULK METALLIC GLASSES. EVOLUTION VERSUS TEMPERATURE IN THE GLASS TRANSITION REGION AND INFLUENCE OF CRYSTALLISATION. B. VAN DE MOORTELE*, J.M. PELLETIER*, J.L. SOUBEYROUX**, I.R. LU*** *GEMPPM, INSA, Bat. 502, INSA, 6962 Villeurbanne Cedex, France. **CRETA, CNRS, Grenoble, France. ***DLR, Köln, Germany. ABSTRACT Three bulk metallic glasses, with different resistance against crystallisation, were investigated using DSC experiments, X-ray diffraction, transmission electron microscopy and mechanical spectroscopy. Like in other non-crystalline materials, the elastic modulus exhibits a large decrease above the glass transition temperature. In materials with a large supercooled region (Pd-Ni-Cu-P for instance), this decrease can reach three decades, leading to an attractive glass forming ability. In contrast, in bulk metallic glasses in which onset of crystallisation occurs very rapidly above T g, this decrease is on less than one decade. A correlation is made with the microstructure evolution revealed by X-Ray diffraction and transmission electron microscopy. INTRODUCTION During the last decade, new multicomponent glass forming alloys have been developed, which exhibit a good glass forming ability [-2]. Some of these bulk metallic glasses (BMG) have a wide supercooled region (SLR), with high resistance against crystallisation. This stability is usually characterised by the parameter Tx = T x T g, where temperature T x corresponds to the onset of crystallisation and T g to glass transition temperature. Values of Tx depend on alloy composition : Pd-Ni-Cu-P BMG exhibit the highest one [8-2]. However, due to the excessive cost of this material, other alloys were developed. Many studies have been performed in Zr- base materials : Zr-Ti-Cu-Ni-Be (vitreloy) [3, 6, 8-0, 5, 7] and Zr-Ti-Cu-Ni-Al (Al replaces the expensive and toxic Be element) [, 2, 7,, 3, 4, 6]. Desirable properties may include mechanical features or high corrosion resistance, for instance for athletic equipment or aerospace components. In many cases, the forming ability is a pertinent characteristics and, consequently, temperature evolution of the mechanical properties has to be investigated (elastic behaviour, viscosity, ). For instance, it is well-known that in non-crystalline materials, elastic modulus exhibits a large decrease above Tg. Unfortunately, in BMG, thermal ability to keep the non-crystalline state is not always ensured, in contrast to polymers or oxide glasses. This thermal stability depends on composition and especially on oxygen content [2]. The present study reports on elastic properties versus temperature of three BMG : Pd-Ni- Cu-P, Zr-Ti-Cu-Ni-Be and Zr-Ti-Cu-Ni-Al. Characteristic temperatures (T g and T x ) were determined by classical DSC experiments and information on microstructure was provided by X-ray diffraction experiments and transmission electron microscopy. L0.6.
EXPERIMENTAL PROCEDURE Ingots of amorphous materials were prepared by induction melting in controlled atmosphere and rapid quenching. Composition, denomination, shape and manufacturer are given in table I for each alloy. Material (at.%) Pd 42.3 Cu 3.3 Ni 0.7 P 5.7 Zr 4 Cu.6 Ni.5 Ti 5.4 Be 20.5 Zr 44.5 Cu 25 Ni 3.5 Ti 5.5 Al Denomination Pd- ZrBe ZrAl Shape Cylinder φ = 5.5 mm Plate 400*250*3.3 mm 3 Cylinder φ = 6 mm Manufacturer I.R. Lu (DLR, Germany) Howmet Corp. (USA) CNRS (Grenoble, France) Table I : Characteristics of the different bulk metallic glasses. A differential scanning calorimetry apparatus (DSC) (Perkin Elmer 7) was used between 200 and 550 C in a purified argon atmosphere. Heating rate were in the range - 60K/min. X-Ray diffraction patterns were obtained by the use of X-ray generator (Cu Kα ). For transmission electron microscopy observations, discs with a diameter of 3 mm and a thickness of 0.5 mm were thinned by PIPS ( GATAN 69). For the investigation a JEOL 200cx was used. The EDX analyze was made on ingots with a scanning electron microscope (JEOL 840ALGS). The shear dynamic modulus G*(ω) was measured by a mechanical spectrometer described by Etienne et al [22]. This apparatus gives data at frequencies between 0-5 Hz and Hz. Rectangular specimens (50*4* mm 3 ) are gripped to the oscillating system. All measurements are performed with a low nitrogen pressure atmosphere. Tan φ is defined as the ratio G (ω)/g (ω), where G and G are the loss and the storage modulus, respectively, and ω the angular frequency. EXPERIMENTAL RESULTS AND DISCUSSION Microstructure evolution Figure shows DSC curves for ZrBe alloy (3, 0 and 20K/min). On each curve there is a small endothermic effect (due to glass transition, see details in fig.2) followed by two exothermic peaks induced by crystallisation (in several stages). Both glass transition and peak crystallisation shift to higher temperature when heating rate is increased. In addition, the relative intensity of the different crystallisation peaks are modified with heating rate. Exothermic (a.u.),4,2 0,6 0,4 20K/min 3K/min 0,2 0K/min 0 370 390 40 430 450 470 490 T( C) Fig. : DSC curve in Zr-Be BMG L0.6.2
In order to compare different glasses, it is important to well define the characteristic temperatures. Fig.2 shows with more details glass transition and supercooled liquid region (SLR) for Pd alloy at a heating rate equal to K/min. It is usual to call Tg and Tg 3 the beginning and the end of the glass transition, respectively. T x is the onset of crystallization. The SLR is defined by T x =T x -T g. The extend of the SLR determine the thermal stability of an amorphous. In our case we use Tg 3 to calculate T x, since it corresponds to the achievement of the glass transition, and, therefore, to a real supercooled liquid. Let us mention that in many reported works, authors give a value of T x without any precision concerning either heating rate or precise definition of the glass temperature value used to calculate this parameter. Consequently, any comparison between different materials is difficult to achieve. Fig. shows clearly that T g and T x change with heating rate. In contrast, the difference T x is only slightly modified by any modification of this rate. In order to study the resistant against crystallization we compare the extend of the SLR for the three alloys. DSC were carried out at the same heating rate (0K/min, see figure 3) and we normalized the C p variations for all the samples. 260 280 300 320 340 360 380 We can see that Pd-alloy is the more stable: indeed the extent of the plateau corresponding to the SLR is about 70K. For ZrAl-alloy the SLR extent is only about 0K. Exothermic (a.u.) Exothermic (a.u.),5 0,5-0,5 -,,05 0,95 0,9 5 Tg Tg3 Tx T ( C) Fig. 2 : definition of the various parameters Pd ZrBe 0 250 300 350 400 450 T( C) ZrAl Fig. 3 : Relative variation of Cp in the three BMG Question is now as follows : are these differences related to any difference in microstructure? X-ray diffraction shows that the three alloy are amorphous, since no indication of crystallization peak is detected. Study in TEM confirms that Pd alloy is fully amorphous. In contrast, in the ZrAl-alloy some crystallites are observed in the bulk (figure 4). The number and the volume fraction (<%) of crystallites are too small to be detected by X-Ray diffraction. The size of these particles is between and 8 µm. Their morphology is complex. µm Fig. 4 : Primary crystallites in Zr-Al alloy Investigations performed at high magnification reveal that they are constituted by a mixture of nanometric amorphous domains and crystals. Holes are often observed at the center L0.6.3
of these partially crystallized zones (figure 4). The role of the hole in the origin of the crystallite is still unknown. The crystalline phase is an unidentified metastable phase. Microanalysis indicates that he crystallite composition seems to be the same as that of the bulk. Determination of the crystallographic structure of this new phase is now in progress. It will be interesting to study the influence of these primary crystallites on the crystallization observed during a further heating. 2 Mechanical properties Fig.5 shows the storage modulus G in the as-cast amorphous Zr-Be alloy, from 20 C up to 600 C. Measurements were performed with a driving frequency of 0.3 Hz and a heating rate of 3 K/min, both during heating and subsequent cooling. Only relative variations are plotted, to enable a better description : G 0 corresponds to the value measured at room temperature in the as-cast BMG. The evolution is schematically as follows : - a slight decrease up to 350 C - a rapid decrease (350-400 C) and a stabilisation (400-450 C) - a rapid increase induced by the onset of crystallisation (first step of decomposition) (450-500 C) - a regular decrease (500-580 C) before the final increase induced by the second stage of crystallisation (580-600 C). G / G 0,6,4,2,0 0,6 0,4 0,2 0,0 0 00 200 300 400 500 600 T ( C) Fig. 5 : Relative variation of the storage modulus in Z-Be BMG - This second stage induces a large modulus increase, as also evidenced by the behaviour during the cooling (performed at the same rate : 3 K/min). Evolution is regular, but all the values are higher than that measured during heating. These irreversible changes yields to a final modulus increase of about 40%. A parallel evolution of tan φ is observed. In order to get a better comparison, evolution of storage modulus and specific heat are also shown in Fig.6. Internal friction (tan φ) is related to the molecular mobility : in the SLR, molecular mobility is very high and, consequently, tan φ is also very high. Onset of crystallisation induces a limitation of mobility and then a decrease of internal friction.,8,6,4,2 0,6 0,4 0,2 0 G /G 0 Cp 300 350 400 450 500 T( C) tan φ Fig. 6 : Internal friction, storage modulus and specific heat vs temperature in Zr-Be BMG Furthermore, an increase of either the heating rate or the testing frequency shifts the characteristic temperatures towards higher values. L0.6.4
Similar results are obtained with the other BMG (Pd- and Zr-Al). In order to compare the influence of material composition, relative temperature (T/T α ) and relative modulus (G /G 0 ) are plotted in Fig. 7. T α represents the onset of the modulus decrease; its value is close to T g in the present testing conditions (0.3 Hz). Therefore we will consider, in the following, that T α = T g. The difference of thermal stability in the,e+0 three BMG is clearly revealed by this representation : Pd- alloy, with the largest SLR, Zr -... - Al,E+00 exhibits the largest decrease (NB : the evolution,e-0 of G during the crystallisation process is not Zr - -Be shown here), while only a limited decrease is,e-02 Pd - - P observed in the Zr-Al BMG (no more than about,e-03 0.3). Zr-Be shows an intermediate behaviour. The consequence of this evolution on forming ability is evident : Pd- BMG can be deformed with a limited applied stress in its wide SLR (without any risk of crystallisation), while in Zr-Al BMG, deformation can be achieved only at temperature close to Tg to prevent this microstructure transformation and the stress level to apply is much larger. This elastic modulus variation vs temperature is now compared to that reported in other non-crystalline materials. The magnitude of the modulus decrease above T g depend on the existence of interbond interactions : - In molecular compounds, elastic modulus tends towards zero at high temperature and then a liquid Newtonian behaviour is observed, characterised by a simple parameter (viscosity). - In polymeric materials, a rubbery plateau is often observed above T g [23]. Then, the storage modulus tends to a finite value G res, even when the α relaxation is fully achieved. This rubbery component originates from the existence of a network of polymeric chains, with physical and chemical cross-links. In thermoplastics (PMMA, for example), the G res /G 0 ratio is typically about 0-4. Existence of stronger cross-links limits this reduction : in elastomers, a typical ratio is about 0-3 and in a thermoset, this value is about 0 - (let us mention that in a fully crosslinked thermoset, like Bakelite, the modulus decrease is even more smaller). Occurrence of crystallisation induces a similar effect : crystallites act like supplementary crosslinks, and, consequently, elastic modulus is increased [24]. - In oxide glasses, situation depends on a lot of parameters : frequency, stress level, composition (especially on the number of bridging oxygen atoms) and a non-linear behaviour in oxide glasses has been reported in a low stress field [25]. So, the elastic behaviour of non-crystalline materials above the glass temperature region depends strongly on interbond interactions (existence of chains, cross-links, crystallites, ). In BMG, the behaviour is in agreement with data reported in other classes of amorphous materials. CONCLUSION 0,9,,2 The elastic properties of three different BMG (PD-Ni-Cu-P, Zr-Ti-Cu-Ni-Be and Zr-Ti- Cu-Ni-Al) were investigated in a large temperature range and especially in the supercooled liquid region, by mechanical spectroscopy. The temperature gap between Tg (glass transition G / G 0 T / Tg Fig. 7 : Relative variations of the storage modulus in the three BMG L0.6.5
temperature) and T g (onset of crystallisation) was first determined by DSC experiments. These temperatures increase with the heating rate and, furthermore, we emphasised the fact that the glass transition spreads over a finite temperature range (typically 20 C).. The storage modulus, i.e. the real part of the complex dynamic shear modulus, exhibits a behaviour similar to that observed in other non-crystalline solids : in materials with a large SLR (especially Pd-Ni-Cu-P and, to a smaller extend, Zr-Ti-Cu-Ni-Be) a drastic decrease is observed (about three decades), leading therefore to a very attractive forming ability. Onset of crystallisation is revealed by a large modulus increase in the supercooled liquid region. REFERENCES - A. Inoue, D. Kawase, A.P. Tsai, T. Zhang and T. Masumoto, Mat. Sci. Eng. A78, p. 255 (994). 2 - A. Inoue, Mater. Sci. Forum 79-8, p. 69 (995). 3 - R. Busch, S. Schneider, A. Peker and W.L. Johnson, Appl. Phys. Letters 67, p. 544, (995). 4 - W.L. Johnson, Mater. Sci. Forum 225-227, p. 35 (996). 5 - Li Ye, R.B. Schwarz, D. Mandrus and L. Jacobson, J. Non Cryst. Sol. 205-207, p. 602 (996). 6 - S. Schneider, P. Thigarajan and W.L. Johnson, Appl. Phys. Letters 68, p. 493 (996). 7 - L.Q. Xing and P. Ochin, Acta Mater. 45, p. 3765 (997). 8 - C.J. Gilbert, R.O. Richie and W.L. Johnson, Appl. Phys. Letters 7, p.476 (977). 9 - J.M. Liu, A. Wiedenmann, U. Gerold, U. Kinderling and H. Wollenberger, Phys. Stat. Sol (b), 99, p. 379 (997) 0 - S. Spriano, C. Antonione, R. Doglione, L. Batteazzati, C. Cardoso, J.C. Soares and M.F. Da Silva, Phil. Mag. B, 76, p. 529 (997). - L.Q. Xing, D.M. Herlach, M. Cornet, C. Bertrand, J.P. Dallas, M.F. Trichet and J.P. Chevallier, Mat. Sci. Eng. A226-228, p. 874 (997). 2 - A. Gebert, J. Eckert and L. Schultz, Acta Mater. 46, p. 5475 (998). 3 - Y. Kawamura, T. Shibata, A. Inoue and T. Masumoto, Acta Mater. 46, p. 253 (998). 4 - L.Q. Xing, C. Bertrand, J.P. Dallas and M. Cornet, Mat. Sci. Eng. A24, P. 26 (998). 5 - W.H. Wang, Q. Wei and S. Friedrich, J. Mat. Sci., 35, p.229 (2000). 6 - J.M. Pelletier, J. Perez and J.L. Soubeyroux, J. Non Cryst. Sol., 274, p.30 (2000). 7 - J.F. Löffler and W.L. Johnson, Appl. Phys. Letters, 76, p.3394 (2000). 8 - J. Löffler, J. Schröers and W.L. Johnson, Appl. Phys. Letters, 77, p. 68 (2000). 9 - I.R. Lu, G.P. Gorler, H.J. Fecht and R. Willnecker, J. Non Cryst. Sol., 274, p. 294 (2000). 20 - J. Schöers, W.L. Johnson and R. Busch, Appl. Phys. Letters, 77, p. 58 (2000). 2 - Y. Kawamura and A. Inoue, Appl. Phys. Letters, 77, p. 4 (2000). 22 - S. Etienne, J.Y. Cavaillé, J. Perez, R. Point and M. Salvia, Rev. Sci. Instr. 53, p. 26 (982). 23 - I.M. Ward and D.W. Hadley, Introduction to the Mechanical Properties of Polymers, Wiley, New-York (993). 24 - C. Gauthier, J.F. Chailan and J. Chauchard, Polymers and Polymer Composites,, p. 309A (993). 25 - J.M. Pelletier, J. Perez, L Duffrene and A. Sekkat, J. Non Cryst. Sol., 258, p. 9 (999). L0.6.6